The final tensile fracture of any composite structure is primarily due to the failure of its constituents, namely fibers and matrix in the present case. To date, no experimental data exists, to the author’s knowledge, to define the behavior of constitutive fibers of a composite structure throughout its life span. The prime candidate for a fiber-based investigation is unidirectional zero-degree composite coupons. But unidirectional coupons do not demonstrate any significant loss of stiffness during fatigue cycling compared to other lay-ups. Even if stiffness degradation was significant, due to the nature of damage in this material system it would be impossible, practically, to monitor that change using conventional techniques (e.g. an extensometer) because the damage and failure process destroys the integrity of the contact between those devices and the material, under cyclic conditions. This investigation presents the findings of a fiber-based investigation of unidirectional composite material systems. In particular, a unidirectional graphite/epoxy system was studied, and the influence of applied load level on fiber fractures, and their influence on damage growth documented. A damage monitoring technique (patent pending) was developed to accurately record the state of damage in this material system without the usage of extensometers or strain gages. Following this method, two new damage norms were introduced, namely, “percent phase damage” and “percent gain damage”. Fiber fracture, strength degradation, and the life of unidirectional specimens were investigated and recorded as a function of various load levels. Fiber fracture, in general, showed no definitive growth pattern during fatigue cycling. It appears that the majority of the broken fibers that occur over nearly 90% of the life are due to the initial applied load cycle. This is one of the key findings of this investigation. “Proof testing” which is a common practice in industry for “verifying” the integrity of a structure, could very well be causing significant subsequent reductions in life. With these findings as a base, it is now possible to postulate the first well-founded mechanistic model of fiber-dominated fatigue degradation under tensile loading. / Ph. D.